ORIGINAL PAPER

Ionic liquid-based dispersive microextraction of nitrotoluenesin water samples

Paula Berton & Bishnu P. Regmi & David A. Spivak &

Isiah M. Warner

Received: 20 December 2013 /Accepted: 12 April 2014# Springer-Verlag Wien 2014

Abstract We describe a method for dispersive liquid-liquidmicroextraction of nitrotoluene-based compounds. This meth-od is based on use of the room temperature ionic liquid (RTIL)1-hexyl-4-methylpyridinium bis(trifluoromethylsulfonyl)imideas the accepting phase, and is shown to workwell for extractionof 4-nitrotoluene, 2,4-dinitrotoluene, and 2,4,6-trinitrotoluene.Separation and subsequent detection of analytes were accom-plished via HPLC with UV detection. Several parameters thatinfluence the efficiency of the extraction were optimized usingexperimental design. In this regard, a Plackett–Burman designwas used for initial screening, followed by a central compositedesign to further optimize the influencing variables. For a 5-mLwater sample, the optimized IL-DLLME procedure requires26 mg of the RTIL as extraction solvent and 680 μL ofmethanol as the dispersant. Under optimum conditions, limitsof detection (LODs) are lower than 1.05 μg L−1. Relative

standard deviations for 6 replicate determinations at a4μg L−1 analyte level are <4.3 % (calculated using peak areas).Correlation coefficients of >0.998 were achieved. This methodwas successfully applied to extraction and determination ofnitrotoluene-based compounds in spiked tap and lake watersamples.

Keywords Room temperature ionic liquid .Microextraction .

Experimental design . Nitroaromatic compounds .Water

Introduction

Nitroaromatic compounds, specifically nitrotoluenes, are re-leased into the environment primarily through waste from thechemical industry, namely, production of pesticides, drugs,explosives, polymers, or dyes, amongst others [1, 2]. Someof these compounds also leach into the soil and eventuallyreach aqueous compartments. Even at low concentrations,nitrotoluenes are very toxic, as well as genotoxic to manyspecies, particularly mammals, for whom evidence of carci-nogenicity has been reported [3–5]. As a result, the USEnvironmental Protection Agency (US EPA) has set a limitof 2 μg L−1 for trinitrotoluene (TNT), and of 0.05 μg L−1 foreach 2,4-dinitrotoluene (2,4-DNT) and 2,6-DNT in drinkingwater [6, 7]. Thus, there is an intense effort to develop effi-cient, sensitive, and cost-effective analytical methods for re-moval and monitoring these pollutants from environmentalcompartments [8]. However, complex matrices and low con-centrations (often at low μg L−1) are a great challenge foraccurate analysis of these species. Therefore, several pretreat-ment steps are required to obtain high enrichment and cleanextracts prior to analyses.

Conventional liquid-liquid extraction (LLE) and solid-phase extraction (SPE) are the most frequently used methodsfor sample pretreatment to isolate and/or enrich nitroaromatic

Electronic supplementary material The online version of this article(doi:10.1007/s00604-014-1261-2) contains supplementary material,which is available to authorized users.

P. Berton :B. P. Regmi :D. A. Spivak : I. M. WarnerDepartment of Chemistry, Louisiana State University, Baton Rouge,LA 70803, USA

P. BertonLaboratory of Analytical Chemistry for Research and Development(QUIANID), Instituto de Ciencias Básicas, Universidad Nacional deCuyo, Padre J. Contreras 1300, Parque Gral, SanMartín M5502JMA, Mendoza, Argentina

P. BertonConsejo Nacional de Investigaciones Científicas y Técnicas(CONICET), Mendoza, Argentina

P. Berton (*)Instituto Argentino de Nivología, Glaciología y CienciasAmbientales (IANIGLA), CCT Mendoza-CONICET, Z.C. 330,5500 Mendoza, Argentinae-mail: pberton@mendoza-conicet.gob.ar

Microchim ActaDOI 10.1007/s00604-014-1261-2

compounds for HPLC determination in aqueous samples [9].However, these techniques have a major disadvantage as theyare often time-consuming and laborious. Furthermore, LLErequires large amounts of high-purity solvents, which areoften volatile, toxic and flammable, and hence result in pro-duction of toxic laboratory waste [10]. More recently, samplepreparation has moved towards environmentally friendly pro-cedures, simplicity, low cost, and automation [10]. In additionto these desirable features, sample preparation must also becharacterized by high accuracy and reliability. Over the lastdecade, a new trend involves miniaturization of LLE proce-dure by greatly reducing the sample-to-solvent ratio. Usingthis approach, dispersive liquid-liquid microextraction(DLLME) has been developed, where the contact area be-tween sample solution and organic solvent is improved byinducing an emulsion of microdroplets after rapid injection ofan appropriate mixture of extraction and dispersant solventsinto the contaminated water sample. DLLME has been suc-cessfully applied for separation and preconcentration of or-ganic and inorganic compounds in environmental samples[11]. Overall, in comparison with other microextraction con-cepts, this recently introduced method represents an attractivealternative because it is rapid, simple, cost-effective, andsensitive [11].

A further improvement in liquid-liquid microextraction(LLME) has been made by replacing organic solvents withroom temperature ionic liquids (RTILs). This substitution re-solves safety and environmental issues associated with the useof organic solvents since RTILs are nonflammable and possesslittle or no detectable vapor pressure [12, 13]. Even thoughRTIL-based methodologies have been extensively used for or-ganic pollutant extraction [12], there is a lack of microextractionmethodologies based on these novel solvents for nitrotoluene-based compound (NC) extraction. In themajority of RTIL-basedLLME, RTILs comprising imidazolium-based cation and thehexafluorophosphate anion have been used [14]. Even thoughRTILs with bis(trifluoromethylsulfonyl)imide (Tf2N) as anionare more stable to hydrolysis, less water soluble, and much lessviscous [15], only a few microextraction studies based on thistype of RTIL has been reported in the literature [16, 17].

In the present work, a rapid and simple cleanup and separa-tion method based on the use of an RTIL is developed for thefirst time for selective and accurate determination of trace levelsof NCs. Using this method, analytes were extracted with 1-hexyl-4-methylpyridinium bis(trifluoromethylsulfonyl)imide([C6mpy][Tf2N]) from pretreated aqueous samples employingthe IL-DLLME technique. This IL-enriched phase was dis-solved in a minimum volume of methanol and subsequentlyused for determination of NCs by use of HPLC. In order toobtain analytical data of the highest quality, and hence toachieve the best system performance, a chemometric experi-mental design was used for data processing. Experimentalvariables were studied by use of a multivariate strategy based

on an experimental design using Plackett-Burman (P-B)screening and a central composite design (CCD) for optimiza-tion of the most influential factors. The optimized procedurewas applied to extraction of NCs in environmental watersamples.

Experimental

Reagents

The compounds 4-picoline (98 %), 1-bromohexane (98 %), 1-methylimidazole (99 %), bis(trifluoromethylsulfonyl)imidelithium salt (LiTf2N, >99 %), 4-nitrotoluene (NT, 99 %),2,6-dinitrotoluene (DNT, 98 %), and sodium chloride (NaCl,≥99.0%) were purchased from Sigma-Aldrich (St Louis, MO,USA, http://www.sigmaaldrich.com). The compound 2,4,6-trinitrotoluene (TNT) was synthetized, purified, and kindlyprovided by Prof. Spivak’s group (Louisiana StateUniversity). Ultrapure water (18.2MΩ cm) was obtained fromMillipore Continental Water System (Bedford, MA, USA,https://www.millipore.com). Stock solutions of 20 g L−1

NCs were prepared in methanol (J.T. Baker, Center Valley,PA, USA, http://jtbaker.com). Lower concentrations wereobtained by diluting the stock solution with ultrapure water.All solutions were stored in the dark at 4 °C. The RTILs 1-hexyl-3-methylimidazolium bis(trifluoromethylsulfonyl)im-ide ([C6mim][Tf2N]) and [C6mpy][Tf2N] were synthesizedusing a method outlined by J.G. Huddleston and coworkers[18]. The synthesized RTILs were characterized by use ofhigh resolution mass spectrometry (HR-MS).

Instrumentation

Separation and quantitative analyses of NCs in aqueous solu-tion were performed on a Shimadzu HPLC system (Kyoto,Japan, http://www.shimadzu.com) consisting of an SCL-10Asystem controller, two LC-10 AD pumps, a DGU-14Adegasser, a SIL-10 AD autosampler and an SPD-10AV UV–vis detector (λ=250 and 275 nm). Separation of analytes wasachieved at room temperature using a Zorbax-CN reverse-phase column (4.6×250 mm, particle size 5 μm), purchasedfrom DuPont Instruments (Wilmington, DE, USA). Analyteswere eluted using an isocratic program 0.1 % trifluoroaceticacid aqueous solution:acetonitrile (40:60) with a flow rate of0.5 mL min−1 for 25 min and monitored at 250 (for DNT andTNT) and 275 nm (for NT). Sample injection volume for thesestudies was 20 μL. Instrumental conditions are summarized inTable S1 in the Electronic Supporting Material (ESM).

Absorbance spectra of stock solutions (background solventmixture: 0.1 % trifluoroacetic acid aqueous solution:acetonitrile(40:60)) were measured in the wavelength range from 200 to600 nm using a Lambda 750 UV/Vis spectrometer

P. Berton et al.

(Perkin Elmer, Shelton, CT, USA, http://www.perkinelmer.com), equipped with 1 cm quartz cuvette.

Sample collection and conditioning

For tap water sample collection, domestic water was allowedto run for 20 min and approximately 1,000 mL were collectedin a beaker. Tap water samples were analyzed immediatelyafter sampling. Lake water samples were collected in Pyrexborosilicate amber glass containers after rinsing three timeswith water sample prior to collection. A sample volume of1,000 mL was collected at a depth of 5 cm below the surface.Once received, samples were immediately filtered through0.45 μm pore size membrane filters (Millipore Corporation,Bedford, MA, USA), due to the high concentration of TDS(total dissolved solids) in these samples. Since clear sampleswere obtained after filtration, they were taken as such forfurther analysis.

Analyses of the samples showed that they were free ofNCs. Water samples were then spiked with NT, DNT, andTNT at two different concentrations, i.e. 5 and 20 μg L−1. Allsamples were stored at 4 °C in brown glass bottles (Nalgene;Nalge, Rochester, NY, USA). All samples were analyzed assoon as possible.

Microextraction procedure

In a 15-mL centrifuge tube, 1.25 g NaCl were added to 5-mLaqueous solution spiked with compounds of interest. Then, asolution of 26 mg of [C6mpy][Tf2N] in 0.68 mL of methanolwas rapidly injected into this aqueous solution, which resultedin the formation of an emulsion. After an extraction time of5 min, the resulting mixture was centrifuged at 2,400 rpm(1,270 g) for 5 min to separate the two immiscible phases. Theupper aqueous phase was discarded, and the lower RTIL-richphase was dissolved in 50 μL of methanol, and injected intothe HPLC for separation and determination of analytes.Calibration was performed by spiking the samples withknown concentrations of the analytes. Environmental samplesand calibration standards followed the same procedures asdescribed above. Optimized instrumental and experimentalconditions are summarized in Table S1 in the ESM.

Results and discussion

Separation conditions

Isocratic elution was suitable for separation of NT, DNT, andTNT in the presence of RTIL dissolved in methanol. Thisseparation was best achieved using a mobile phase containing40 % of 0.1 % trifluoroacetic acid prepared in high puritywater (mobile phase A) and 60% of acetonitrile (mobile phase

B) at a flow rate of 0.5 mL min−1. The selected column,Zorbax CN, has a cyano-bonded phase design that can beused for many of the same separations as in a C18 columnwhile avoiding some of the disadvantages of C18 columnssuch as poor wettability in high aqueous mobile phases whichare generally employed during separation of NCs. Under theseconditions, NT, DNT, and TNT were eluted within 25 min,and no significant interfering peaks from the RTIL wereobserved at their respective retention times.

RTIL selection

The selection of a suitable RTIL was achieved by use ofspecific properties required for IL-DLLME, including lowsolubility in water, good extraction ability, low viscosity, andhigher density than water. While a lower solubility allowsminimal RTIL consumption, a high viscosity could lead topractical problems during microextraction. In addition, ahigher concentration of RTIL in the aqueous phase coulddecrease the partition coefficient of the extracting phase foranalytes. Therefore, RTILs which are relatively easy to syn-thesize, hydrophobic and with high-density were studied(properties shown in Table S2 in the ESM). RTILs withTf2N anion are usually less viscous, while those withpyridinium cation exhibit lower solubility in water.Therefore, [C6mim][Tf2N] and [C6mpy][Tf2N] were selectedas possible extraction solvents for this microextractiontechnique.

Selection of microextraction type

Various strategies for inducing emulsions during the DLLMEprocedure were considered in order to achieve maximumextraction recovery of analytes. Accordingly, IL-basedDLLME (IL-DLLME), temperature-controlled IL dispersiveliquid phase microextraction (TILDLME), ultrasound-assisted IL-DLLME (IL-USA-DLLME), and various combi-nations of these were investigated. These approaches haveseveral advantages in common including simplicity, rapidity,and cost effectiveness [13]. In order to use temperature-basedmicroextraction, the stability of these compounds at differenttemperatures (from 60 to 80 °C) and times (5 to 20 min) wereevaluated. These temperatures and times were selected basedon requirements for solubilization of the RTIL phase into theaqueous phase during TILDLME processes. It was observed(data not shown) that analytes, particularly NT, are unstable,even for short exposures at relatively low temperatures such as60 °C. Therefore, temperature-based microextractions werenot considered. Other dispersive microextraction techniquessuch as IL-DLLME and IL-USA-DLLME were then testedfor extraction of NCs. Classical IL-DLLME is based on aternary solvent system in which a third disperser solvent isused to induce emulsion formation. In IL-USA-DLLME, an

Ionic liquid-based dispersive microextraction of nitrotoluenes

ultrasonic bath is used to induce the formation of a cloudysolution of extraction solvent into an aqueous sample. Asshown in Fig. 1, an improvement on extraction of analytes isobserved when IL-DLLME is used (p-values <0.05). In goodagreement with Liu et al. [19], chemical dispersion was foundto be more effective than physical dispersion when RTILswere used as extractant phases with analytes present in watersamples. The addition of a disperser solvent decreases inter-facial tension between the two phases and accelerates theformation of droplets in water samples, thus increasing thesurface area for extraction of target analytes [20].

Non-statistical differences (p-values >0.05) were found whencomparing extraction recoveries achieved with [C6mpy][Tf2N]and [C6mim][Tf2N] using IL-DLLME. However, the signifi-cantly lower solubility of [C6mpy][Tf2N] in water (Table S2 inthe ESM) allows the use of lower amounts of the RTIL duringthemicroextraction procedure. Therefore, a dispersive techniquebased on [C6mpy][Tf2N], employing a third solvent, was select-ed for further experiments.

Experimental design

In general, single-factor procedures are usually performed tooptimize DLLME parameters relevant to the extraction pro-cess. However, univariate designs are laborious, tedious, nor-mally require a large number of experiments, and do notevaluate the interactive effects of the tested variables [21].To overcome these problems, a chemometric design wasdeveloped for optimization of our method, considering itsadvantages with regard to the requirement of less resource(time, reagents, experimental work) [22]. Due to the largenumber of parameters to be tested, the well-known P-B design(two-level fractional design) was used to identify those pa-rameters that may significantly impact the responses. After

evaluating the data and statistically significant effects throughthe analysis of variance test, significant variables were select-ed using a Pareto chart [23]. Then, a response surface meth-odology, termed CCD, was used to study the selected vari-ables and their interactions more extensively. After experi-mental procedures, outliers were removed by analyzing dif-ferences amongst fitted values test (DFFITS) due to a dispro-portionate influence on the predicted response, and thus on themodel [23]. After a mathematical model was obtained for eachanalyte to describe the response behavior in the experimentaldomain, the conditions for simultaneous extraction of allanalytes were found by use of a multiple response optimiza-tion technique employing the desirability function (D) [24].Experimental design, data analysis, and desirability functioncalculations were performed by use of the software Stat-EaseDesign-Expert Version 8.0.7.1 (2011) (Stat-Ease Inc.,Minneapolis).

Screening design

In order to obtain high extraction recoveries, a P-B design wasused for screening and optimization of the effective factors for

0

20

40

60

80

100IL-DLLME IL-USA-DLLME

[C6mim]A[C

6mim]A [C

6mpy]A[C

6mpy]A

)%(

yrevocernoitcartx

E

Fig. 1 Effect of different dispersive-based microextraction techniques(top) and of two RTILs ([C6mim]A and [C6pym]A; A: Tf2N) (down) onextraction recoveries of ( ) NT, ( ) DNT, and (■) TNT. Other experi-mental conditions are illustrated in Table S1 in the ESM

25.0 36.3

47.5 58.8

70.0 0.250.38

0.500.62

0.75

0.00

0.24

0.48

0.72

0.96

Des

irab

ility

RTIL mass(mg)

Disperser solventvolume (mL)

(a)

0.56.6

12.818.9

25.0

0.25 0.38

0.50 0.62

0.75

0.00

0.24

0.48

0.72

0.96

Des

irab

ility

NaCl concentration(% w/v)

(b)

Disperser solventvolume (mL)

Fig. 2 Response surface plots corresponding to the D function whenoptimizing: a RTIL mass vs. disperser solvent volume; b NaCl concen-tration vs. disperser solvent volume

P. Berton et al.

efficient extraction using the IL-DLLME method. The select-ed factors (low factor—high value) were (A) pH (4–8); (B)buffer concentration (5–15 mmol L−1); (C) ionic strength(0–4 % (w/v) NaCl); (D) RTIL mass (25–50 mg); (E) dispers-er solvent (methanol—acetone); (F) disperser solvent volume(0.3–0.5 mL); (G) ultrasound assisted (no—yes); (H) extrac-tion time (5–25 min); (J) centrifugation time (5–15 min); and(K) centrifugation speed (1,400–2,400 rpm (430–1,270 g).Selection of low and high values of the selected variableswere based on prior knowledge of the system under studyand preliminary assays [22]. Normalized results of the exper-imental design were evaluated at 5 % of significance, andanalyzed and visualized by use of standardized main effectPareto charts (shown in Figure S1 in the ESM). Results ofscreening design showed that parameters that primarily influ-enced the recovery extraction of analytes were ionic strength(C), RTIL mass (D), type of dispersant solvent (E), volume ofdispersant solvent (F), and centrifugation speed (K). Based onthese observations, these variables were selected and furtherconsidered for CCD analysis, with the exception of centrifu-gation speed. Even when better recoveries were observed withhigher centrifugation speeds, velocities higher than 2,400 rpmmay compromise the structural integrity and lifetime of cen-trifugation tubes. Therefore, 2,400 rpm was selected for fur-ther experiments.

As expected, the pH of the solution and buffer concentra-tion did not have a significant influence on the extractionrecovery since the analyzed NCs are all non-ionizable inaqueous solutions, due to their high pKa values (NT: 11.35;DNT: 13.75; and TNT: 16.04) [25, 26]. With regard to thedispersant solvent, its selection was based on its miscibility inboth the RTIL and aqueous phase. The highest extractionrecovery was obtained when methanol was used.Application of ultrasound energy is sometimes used to im-prove the extraction efficiency attained with dispersive-basedmicroextraction techniques, due to formation of smaller drop-let sizes of the dispersed RTIL phase upon application ofultrasound energy [27]. However, in the present study, theextraction recovery was not enhancedwhen ultrasound energywas applied to the samples. Similarly, extraction time had nosignificant effect on the extraction efficiency. This is the mostimportant advantage of DLLME, i.e. showing that it is time-independent due to the infinitely large surface area betweenthe extraction solvent and the aqueous phase in the emulsion[20]. Hence, in order to keep the procedure simple and rapid,both extraction and centrifugation times were set at 5 min,with no pH adjustment before extraction and without use ofultrasound energy during extraction.

Response surface methodology

After choosing the effective parameters, a CCD was per-formed to estimate values of the previously selected variablesin order to obtain maximum recovery of analytes. This designallowed responses to be fitted to a polynomial model, whichconsisted of 20 experiments based on combinations of select-ed variables within the following ranges: (a) NaCl concentra-tion: 0.50–25.0 % (w/v); (b) RTIL mass: 25–75 mg; and (c)dispersant solvent volume: 0.25–0.75 mL. Other variablessuch as pH, buffer concentration, dispersant solvent type,ultrasound use, extraction time, and centrifugation time andspeed were set according to results obtained in the screeningphase. The values obtained were 1) 5 mL of sample, 2)methanol as dispersant solvent, 3) 5 min as extraction time,and 4) 5 min of centrifugation at 2,400 rpm (1,270 g).

Table 1 Analytical parameters obtained for the present method

Parameter NT DNT TNT

Calibration range (µg L−1) 3–40

Correlation coefficients (r) 0.9986 0.9992 0.9990

LOD (µg L−1) 1.05 0.70 0.81

RSD (%) 4.32 3.09 3.42

Retention time (min) 17.7 18.5 21.3

Capacity factor (k’) 1.68 1.83 2.25

Efficiency factor (N) 5,550.4 2,049.3 3,065.0

Separation factor (α) 1.09 1.23

5.482.72Resolution (R)

Table 2 Comparison of thepresent method with othermicroextraction-based methodsreported for the determination ofNCs in water

a Not reported

Technique Extractiontime (min)

LOD(μg L−1)

RSD (%) Sampleconsumption (mL)

Calibrationrange (μg L−1)

Ref

SPME 65 1.00–10.0 10.2–27.2 25 2–200 [31]

SPME 62 0.17–0.92 1.5–3.5 20 10–400 [32]

SPME 30 0.60–120 1.2–4.8 6 a [33]

EA D-μSPE 5 1.80–7.00 1.8–8.6 10 15–1,000 [30]

D-μSPE 2 0.11–1.26 4.4–9.5 200 4.1–50 [34]

DLLME 2 0.09–0.5 a 9 0.5–400 [28]

IL-DLLME 5 0.70–1.05 3.09–4.32 5 3–40 Present work

Ionic liquid-based dispersive microextraction of nitrotoluenes

Data obtained were evaluated by use of an ANOVAtest. The resulting behavior of the analytical response ofthe analyzed nitroaromatic compounds under the influ-ence of the studied variables was best explained by useof quadratic models. Statistical parameters correspondingto the fitting for resolution demonstrated the significanceof the models (p<0.05) and the non-significance of thelacks of fit (p>0.05). Evaluation of p-values at 95 %confidence showed that NaCl concentration and volumeof dispersant solvent significantly impact the extractionrecovery of the target analytes and thus their analyticalresponse. Depending on the contaminants and the selectedextraction technique, addition of salt to the sample solu-tion can decrease solubility. This is due to a salting outeffect, where water molecules tend to hydrate the saltions, and thus are less available for dissolution of analytemolecules. In this case, increasing the ionic strength pro-moted transport of analytes into the RTIL phase.Regarding dispersant solvent volume, an increase in ex-traction recovery was observed with higher volumes ofsolvent. Higher volumes were not evaluated since it waspreviously observed that non-significant to negative ef-fects were observed when higher volumes were employed[28].

Optimization of IL-DLLME technique

After CCD, the D function was used to define the valuesthat lead to best analytical response for the studied NCs.The D function also includes researcher’s priorities anddesires on building the optimization procedure. The pro-cedure creates a function for each individual analyticalresponse, and finally obtains a global D function thatshould be maximized by choice of the best conditionsfor design variables. In order to prioritize the concept ofgreen chemistry, in addition to maximum extraction ofanalytes, another desired condition was to use the mini-mum amount of RTIL, since both the extraction recoveryand analyte detection in HPLC can be remarkably affectedby amount of RTIL.

Following the above conditions and restrictions, theoptimization procedure was then performed and responsesurfaces obtained for global D are exemplified in Fig. 2.These plots were obtained for a given pair of variables,while maintaining the other fixed variable at its optimalvalue. This figure shows two surfaces as example: disper-sant solvent volume vs. RTIL amount (Fig. 2a); anddispersant solvent volume vs. NaCl concentration(Fig. 2b).

Under the optimization criteria noted above, experi-mental conditions corresponding to one of maximum Dfunction (D=0.950) are shown in Table S1 in the ESM.Values suggested by use of optimization procedures werealso experimentally corroborated.

Analytical performance

Several analytical parameters, summarized in Table 1,were calculated to characterize the microextraction proce-dure. Calibration curves were determined using six work-ing aqueous standards in the concentration range between3 μg L−1 and up to at least 40 μg L−1 for the studiedanalytes, with correlation coefficients (r) higher than0.998. The limits of detection (LODs), which were calcu-lated based on the signal at the intercept and three timesthe standard deviation about regression of the calibrationcurve, varied between 0.7 μg L−1 (DNT) and 1.05 μg L−1

(NT). Reproducibility of the method was evaluated at4 μg L−1 (n=6) with results (expressed as relative stan-dard deviation, or RSD) in the range from 3.1 % (DNT) to4.3 % (NT). Chromatographic parameters such as capacityfactor (kA), separation factor (α), efficiency factor (N),and resolution equation (R) were also calculated to eval-uate the methodology (Table 1). Moreover, reproducibleretention times were observed throughout a normal work-ing day (8–12 h of analysis).

The microextraction technique was evaluated in com-parison with others described for determination of NCs inwaters (Table 2) . Amongs t these , so l id phasemicroextraction (SPME) allows a wide calibration range

Table 3 Determination of NT,DNT, and TNT in water samples(95 % confidence interval; n=3)

Analyte Added (μg L−1)a Tap water Lake water

Found (μg L−1) Recovery (%) Found (μg L−1) Recovery (%)

NT 5.00 4.78±0.19 95.7 4.66±0.20 93.2

20.0 19.4±0.68 96.8 19.4±0.74 96.9

DNT 5.00 4.90±0.15 98.1 4.89±0.15 97.8

20.0 20.0±0.54 99.8 19.4±0.57 97.2

TNT 5.00 5.07±0.17 101 4.94±0.17 98.9

20.0 19.9±0.66 99.7 19.5±0.62 97.5

P. Berton et al.

using relatively low volumes of sample. However, theSPME procedure requires more than 30 min instead ofthe 10 min needed in this IL-DLLME approach.Furthermore, the lifetime of fibers used for SPME islimited as it degrades with increased usage, resulting inpeaks that may co-elute with the target analytes [29]. TheLODs achieved with the present method are comparablewith those reported for dispersive micro-SPE (D-μSPE)approaches, with a lower consumption of sample. In ad-dition, effervescence-assisted D-μSPE (EA D-μSPE) re-quires preparation of tablet (ca. 1 h), which should bemaintained under an inert atmosphere until use, which canbe considered a limitation of the method [30].

Application of the methodology to real water samples

In order to evaluate the effect of the matrix on the presentmethodology, this procedure was applied to analyses of watersamples. Initial analysis of these samples revealed that theywere free of tested compounds. Then, a recovery test wasperformed by spiking the water samples with the threeanalytes at two different concentrations (5 and 20 μg L−1),

each being analyzed in triplicate. As seen in Table 3, theconcentration of the nitroaromatic explosives determined inthe spiked water samples agreed reasonably well with theknown values. The average relative recoveries (avr) rangesfrom 95.7 to 101 % for tap water, while for lake watersamples, the avr ranged from 93.2 to 98.9 %, demonstratinglittle effect of the matrix on the analysis of these samples. Arepresentative chromatogram of water after IL-DLLME-HPLC analysis is shown in Fig. 3.

The high reproducibility of retention times for NT, DNT,and TNT obtained in the presence of the sample matrix,together with the lack of interfering peaks at the retentiontimes of the analytes, may allow application of the methodto other liquid samples.

Conclusions

In this study, a dispersive microextraction based on an RTILcoupled to HPLC is presented as a viable technique for quan-tification of trace levels of NCs in water samples. Through theuse of a multivariate optimization strategy, a successful deter-mination of optimal IL-DLLME conditions was achieved,with a minimum number of analytical assays. The extractionrecoveries of NCs were significantly influenced by NaClconcentration, mass of RTIL, and the dispersant solvent typeand volume.

Evaluation of the results shown in this work indicatestha t the procedure i s s imple and rap id forpreconcentration and determination of NCs in watersamples. Furthermore, since solvent microextractionsdo not need dedicated and expensive apparatus, thedeveloped method is simple and cost-effective whencompared to other microextraction methodologies basedon solid phase extraction procedures. Other advantagesof this method are negligible volumes of solvents usedfor recovery, while using low cost and widely availableinstrumentation. The methodology shows a lack of in-terfering peaks, and specifically for TNT, an LOD valuewell below accepted drinking water standards and healthadvisory figures established by the US EPA. Further, thesubstitution of volatile organic solvents such as carbontetrachloride by ionic liquid as extracting solvents rep-resents a step forward to improve safety in laboratories.Therefore, the present microextraction method based onRTILs is a powerful alternative to conventional extrac-tion methods for analyses of NCs in water samples.

Acknowledgments This material is based upon work supported in partby the National Science Foundation under Grant Numbers CHE-1243916and CHE-1307611; and funds from the Philip W. West Endowment toIMW. A Fulbright & Bunge and Born foundation fellowship granted bythe Council for International Exchange of Scholars, USA, which support-ed Paula Berton is also gratefully acknowledged.

16 18 20 22 240

50

100

150

200(a)

NT

TNTDNT

Abs

orba

nce

(mA

U)

Time (min)

16 18 20 22 240

50

100

150

200

Abs

orba

nce

(mA

U)

Time (min)

(b)

NT

DNT

TNT

Fig. 3 Chromatogram obtained in lake water samples spiked at 5 μg L−1

of each analyte after IL-DLLME-HPLC method for: a DNT and TNT(250 nm); b NT (275 nm). Experimental conditions are illustrated inTable S1 in the ESM

Ionic liquid-based dispersive microextraction of nitrotoluenes

References

1. Keith L, Telliard W (1979) ES&T special report: Priority pollutants:I-a perspective view. Environ Sci Technol 13(4):416–423

2. Ju KS, Parales RE (2010) Nitroaromatic compounds, from synthesisto biodegradation. Microbiol Mol Biol Rev 74(2):250–272

3. Office of Water (2012) 2012 edition of the drinking water standardsand health advisories. U.S. Environmental Protection Agency,Washington, DC

4. Maeda T, Nakamura R, Kadokami K, Ogawa HI (2007) Relationshipbetween mutagenicity and reactivity or biodegradability fornitroaromatic compounds. Environ Toxicol Chem 26(2):237–241

5. Chandra SA, Nolan MW, Malarkey DE (2010) Chemical carcinogen-esis of the gastrointestinal tract in rodents: an overview with emphasison NTP carcinogenesis bioassays. Toxicol Pathol 38(1):188–197

6. Agency for Toxic Substances and Disease Registry (ATSDR) (1995)Toxicological profile for 2,4,6-trinitrotoluene (TNT) (trans: U.S.Department of Health and Human Services). Public Health Service,Atlanta

7. U.S. Environmental Protection Agency (2008) Drinking water healthadvisory for 2,4-dinitrotoluene and 2,6-dinitrotoluene (trans: Healthand Ecological Criteria Division). Washington, DC

8. Kulkarni M, Chaudhari A (2007) Microbial remediation of nitro-aromatic compounds: an overview. J Environ Manag 85(2):496–512

9. Zahedi M, Rahimi-Nasrabadi M, Pourmortazavi S, Koohbijari G,Shamsi J, Payravi M (2012) Emulsification-based dispersive liquidmicroextraction and HPLC determination of carbazole-based explo-sives. Microchim Acta 179(1–2):57–64

10. Pena-Pereira F, Lavilla I, Bendicho C (2010) Liquid-phasemicroextraction techniques within the framework of green chemistry.TrAC Trends Anal Chem 29(7):617–628

11. Han D, RowK (2012) Trends in liquid-phase microextraction, and itsapplication to environmental and biological samples. MicrochimActa 176(1–2):1–22

12. Ma J, Hong X (2012) Application of ionic liquids in organic pollut-ants control. J Environ Manag 99:104–109

13. Aguilera-Herrador E, Lucena R, Cárdenas S, Valcarcel M (2011)Sample treatments based on ionic liquids. In: Kokorin A (ed) Ionicliquids: applications and perspectives. InTech, Croacia

14. Zgoła-Grześkowiak A, Grześkowiak T (2011) Dispersive liquid-liquid microextraction. TrAC Trends Anal Chem 30(9):1382–1399

15. Freire MG, Neves CMSS, Carvalho PJ, Gardas RL, Fernandes AM,Marrucho IM, Santos LMNBF, Coutinho JAP (2007) Mutual solu-bilities of water and hydrophobic ionic liquids. J Phys Chem B111(45):13082–13089

16. Molaakbari E, Mostafavi A, Afzali D (2011) Ionic liquid ultrasoundassisted dispersive liquid–liquid microextraction method forpreconcentration of trace amounts of rhodium prior to flame atomicabsorption spectrometry determination. J Hazard Mater 185(2–3):647–652

17. Martinis EM, Berton P, Monasterio RP, Wuilloud RG (2010)Emerging ionic liquid-based techniques for total-metal and metal-speciation analysis. TrAC Trends Anal Chem 29(10):1184–1201

18. Huddleston JG, Willauer HD, Swatloski RP, Visser AE, Rogers RD(1998) Room temperature ionic liquids as novel media for ‘clean’liquid-liquid extraction. Chem Commun 16:1765–1766

19. Liu Y, Zhao E, Zhu W, Gao H, Zhou Z (2009) Determination offour heterocyclic insecticides by ionic liquid dispersive liquid–liquid microextraction in water samples. J Chromatogr A1216(6):885–891

20. Rahimi-NasrabadiM, ZahediM, Pourmortazavi S, Heydari R, Rai H,Jazayeri J, Javidan A (2012) Simultaneous determination ofcarbazole-based explosives in environmental waters by dispersiveliquid—liquid microextraction coupled to HPLC with UV–vis detec-tion. Microchim Acta 177(1–2):145–152

21. Herrera-Herrera AV, Asensio-Ramos M, Hernández-Borges J,Rodríguez-Delgado MÁ (2010) Dispersive liquid-liquidmicroextraction for determination of organic analytes. TrAC TrendsAnal Chem 29(7):728–751

22. Stalikas C, Fiamegos Y, Sakkas V, Albanis T (2009) Developmentson chemometric approaches to optimize and evaluatemicroextraction. J Chromatogr A 1216(2):175–189

23. Montgomery DC (2001) Design and analysis of experiments. Wiley,New York

24. Myers RH, Montgomery DC (1995) Response surface methodology:process and product optimization using designed experiments, 1stedn. Wiley, New York

25. Guan W, Xu F, Liu W, Zhao J, Guan Y (2007) A newpoly(phthalazine ether sulfone ketone)-coated fiber for solid-phasemicroextraction to determine nitroaromatic explosives in aqueoussamples. J Chromatogr A 1147(1):59–65

26. Eley DD (1991) Advances in catalysis, vol 37. Elsevier Science, SanDiego

27. Priego Capote F, Luque de CastroMD (2007) Analytical applicationsof ultrasound, vol 26, 1st edn, Techniques and instrumentation inanalytical chemistry. Elsevier, Amsterdam

28. Ebrahimzadeh H, Yamini Y, Kamarei F (2009) Optimization ofdispersive liquid–liquid microextraction combined with gas chroma-tography for the analysis of nitroaromatic compounds in water.Talanta 79(5):1472–1477

29. Psillakis E, Kalogerakis N (2001) Solid-phase microextraction versussingle-drop microextraction for the analysis of nitroaromatic explo-sives in water samples. J Chromatogr A 938(1–2):113–120

30. Lasarte-Aragonés G, Lucena R, Cárdenas S, Valcárcel M (2011)Effervescence-assisted dispersive micro-solid phase extraction. JChromatogr A 1218(51):9128–9134

31. Monteil-Rivera F, Beaulieu C, Deschamps S, Paquet L, Hawari J(2004) Determination of explosives in environmental water samplesby solid-phase microextraction–liquid chromatography. JChromatogr A 1048(2):213–221

32. Gaurav, Malik AK, Rai PK (2009) Development of a new SPME–HPLC–UV method for the analysis of nitro explosives on reversephase amide column and application to analysis of aqueous samples.J Hazard Mater 172(2–3):1652–1658

33. Furton KG, Almirall JR, Bi M, Wang J, Wu L (2000) Application ofsolid-phase microextraction to the recovery of explosives and ignit-able liquid residues from forensic specimens. J Chromatogr A 885(1–2):419–432

34. Reyes-Gallardo EM, Lasarte-Aragones G, Lucena R, Cardenas S,Valcarcel M (2013) Hybridization of commercial polymeric micro-particles and magnetic nanoparticles for the dispersive micro-solidphase extraction of nitroaromatic hydrocarbons from water. JChromatogr A 1271(1):50–55

P. Berton et al.